You are on page 1of 15

IOP PUBLISHING J. Neural Eng.

6 (2009) 016001 (15pp)

JOURNAL OF NEURAL ENGINEERING

doi:10.1088/1741-2560/6/1/016001

Creation of highly aligned electrospun poly-L-lactic acid bers for nerve regeneration applications
Han Bing Wang1, Michael E Mullins1, Jared M Cregg2, Andres Hurtado3,4, Martin Oudega3,4, Matthew T Trombley2 and Ryan J Gilbert2
Department of Chemical Engineering, Michigan Technological University, Houghton, MI 49931, USA Regeneration and Repair Laboratory, Department of Biomedical Engineering, Michigan Technological University, Houghton, MI 49931, USA 3 International Center for Spinal Cord Injury, Hugo W Moser Research Institute at Kennedy Krieger, Baltimore, MD 21205, USA 4 Department of Neurology, Johns Hopkins University School of Medicine, Baltimore, MD 21205, USA
2 1

E-mail: rgilbert@mtu.edu

Received 25 November 2008 Accepted for publication 25 November 2008 Published 22 December 2008 Online at stacks.iop.org/JNE/6/016001 Abstract Aligned, electrospun polymer bers have shown considerable promise in directing regenerating axons in vitro and in vivo. However, in several studies, nal electrospinning parameters are presented for producing aligned ber scaffolds, and alignment where minimal ber crossing occurs is not achieved. Highly aligned species are necessary for neural tissue engineering applications to ensure that axonal extension occurs through a regenerating environment efciently. Axonal outgrowth on bers that deviate from the natural axis of growth may delay axonal extension from one end of a scaffold to the other. Therefore, producing aligned ber scaffolds with little ber crossing is essential. In this study, the contributions of four electrospinning parameters (collection disk rotation speed, needle size, needle tip shape and syringe pump ow rate) were investigated thoroughly with the goal of nding parameters to obtain highly aligned electrospun bers made from poly-L-lactic acid (PLLA). Using an 8 wt% PLLA solution in chloroform, a collection disk rotation speed of 1000 revolutions per minute (rpm), a 22 gauge, sharp-tip needle and a syringe pump rate of 2 ml h1 produced highly aligned ber (1.21.6 m in diameter) scaffolds veried using a fast Fourier transform and a ber alignment quantication technique. Additionally, the application of an insulating sheath around the needle tip improved the rate of ber deposition (electrospinning efciency). Optimized scaffolds were then evaluated in vitro using embryonic stage nine (E9) chick dorsal root ganglia (DRGs) and rat Schwann cells (SCs). To demonstrate the importance of creating highly aligned scaffolds to direct neurite outgrowth, scaffolds were created that contained crossing bers. Neurites on these scaffolds were directed down the axis of the aligned bers, but neurites also grew along the crossed bers. At times, these crossed bers even stopped further axonal extension. Highly aligned PLLA bers generated under optimized electrospinning conditions guided neurite and SC growth along the aligned bers. Schwann cells demonstrated the bipolar phenotype seen along the bers. Using a novel technique to determine ber density, an increase in ber density correlated to an increase in the number of neurites, but average neurite length was not statistically different between the two different ber densities. Together, this work presents methods by which to produce highly

1741-2560/09/016001+15$30.00

2009 IOP Publishing Ltd Printed in the UK

J. Neural Eng. 6 (2009) 016001

H B Wang et al

aligned ber scaffolds efciently and techniques for assessing neurite outgrowth on different ber scaffolds, while suggesting that crossing bers may be detrimental in fostering efcient, directed axonal outgrowth. (Some gures in this article are in colour only in the electronic version)

1. Introduction
The development of novel scaffolds able to guide axonal growth is important to facilitate axonal regeneration through injured environments in the peripheral nervous system (PNS) and central nervous system (CNS). The material should be biocompatible and closely resemble native extracellular matrix environments that neurons use to guide axonal extensions during development; it should also be degradable so that the scaffold degrades as axons grow through the injury site. The mechanical properties of the material are also an important parameter as tough, rigid materials may induce additional damage following implantation and/or make it difcult for regenerating axons to penetrate into the material (Balgude et al 2001) Although the desired properties of a neural scaffold, such as biodegradability, biocompatibility and biomechanical character, have been extensively studied, creating a system that satises all such properties remains elusive. Recently, others have used novel fabrication techniques to design materials that guide axonal outgrowth. Aligned conduits (Hadlock et al 2000, Stokols and Tuszynski 2006), bers/laments (Chauhan et al 1999, Rangappa et al 2000, Ngo et al 2003, Cai et al 2005, Wen et al 2006, Schnell et al 2007, Corey et al 2007, 2008, Kim et al 2008) and hydrogels (Ceballos et al 1999, Luo et al 2004, Prang et al 2006, Dodla and Bellamkonda 2006, 2008) which facilitate directed growth of axons in vitro and in vivo have all shown promise in assisting axonal extension in a directed manner. However, there is no standard material or fabrication technique that is generally accepted as being optimum for facilitating directional outgrowth. Thus, further development of novel scaffolds is necessary. Electrospinning is increasingly being investigated to create scaffolds for tissue engineering applications. In neural tissue engineering, the longitudinal distribution of axons in both the CNS and the PNS makes it imperative to design scaffolds able to guide regenerating axons along their natural axis of growth. To date, in vitro experiments using rat DRGs on PLLA (Corey et al 2007) or poly(acrylonitrileco-methylacrylate (Kim et al 2008), chick DRGs on poly-caprolactone (Schnell et al 2006), neural stem cells on PLLA (Yang et al 2004, 2005) and primary motor neurons (from spinal cord) and sensory neurons (from DRGs) from rat on PLLA (Corey et al 2008) have demonstrated the ability of aligned bers to facilitate directed neurite outgrowth in comparison to randomly oriented ber species. Further, aligned, electrospun bers have shown promise in fostering robust regeneration in vivo within a rat peripheral nerve injury model either without neurotrophin (Kim et al 2008) or with
2

neurotrophin (Chew et al 2007). While bers in these studies were more aligned than randomly oriented bers, a number of bers still crossed. Crossing bers are signicant since they may restrict or divert axonal outgrowth from occurring down the longitudinal axis of the implant. Therefore, manipulation of electrospinning parameters is essential to produce highly aligned ber species that facilitate nerve regeneration through an injury site as quickly and as efciently as possible. In this study, electrospinning parameters have been manipulated to produce highly aligned ber specimens by manipulating four electrospinning parameters: the rotation speed of the collection disk, the needle size, the needle tip shape and the syringe pump ow rate. A novel insulating sheath was introduced to the needle tip to increase electrospinning efciency. Samples were imaged using scanning electron microscopy (SEM) and the ber density and alignment were characterized. A novel procedure for determining ber density was established. E9 chick DRG explants and rat SCs were cultured on optimized scaffolds that contained aligned bers at two different densities, respectively. By manipulating electrospinning conditions to produce bers with little to no crossing, highly aligned ber scaffolds directed neurite outgrowth parallel to the bers and neurites growing perpendicular to the orientation of the bers did not occur. The SCs also attached on and grew along the PLLA bers. In comparison, DRGs grown on ber specimens that contained crossed bers grew along the axis of most of the bers. However, several did also grow along the crossed bers, and some axons were impeded by the crossed bers. Using a novel technique to quantify neurite density, it was observed that increasing PLLA ber density correlated to an increase in neurite density without affecting the length of extending neurites.

2. Materials and methods


2.1. Preparation of aligned PLLA electrospun bers PLLA was chosen in this study due to its biocompatibility and biodegradability; 8 wt% PLLA (NatureWorksTM ; grade 6201D, Lot #9051-89-2, density: 1.25, weight average MW being 78 kDa and number average MW being 48 kDa, provided by Cargill Dow LLC, Minnetonka, MN) was dissolved in chloroform at room temperature. A modied electrospinning method (Xu et al 2004) was used to create aligned PLLA bers as shown in gure 1. A high-voltage power supply (Gamma High Voltage Research; Ormond Beach, FL) was used to supply the required charge. A multi-speed syringe pump (Braintree Scientic Inc.; Braintree, MA) was placed perpendicularly to the ground. A 5.0 ml glass syringe and

J. Neural Eng. 6 (2009) 016001

H B Wang et al

Table 1. Electrospinning parameters varied in the optimization tests. Variables Rotation speed of the collector Needle tip shape and gauge Insulated needle Flow rate of the syringe pump 250 rpm 500 rpm 1000 rpm Flat tip Sharp tip Insulated needle 2 mL h1 4 mL h1 Fixed parameters 22 G sharp-tip needle, 4 mL h1, 1 h 20 G needle, 1000 rpm, 4 mL h1, 1 h 22 G sharp-tip needle, 1000 rpm, 4 mL h1, 1 h 22 G sharp-tip needle, 1000 rpm, 1 h

Figure 1. Schematic of the setup to create aligned PLLA bers by electrospinning using an insulated sharp needle and a rotating disk collector.

Figure 2. Alignment analysis. The reference line (Lr1) represents the uniaxial orientation of most of the bers. The angle formed between 150 bers and Lr1 was measured and averaged.

2.2. Scanning electron microscopy one of several needles (20G at and sharp tip needles with an inner diameter of 0.6 mm, and a 22G sharp-tip needle with an inner diameter of 0.4 mm (Fisher Chemicals; Fair Lawn, NJ)) were used to generate bers. The needle was connected to the power supply to charge the polymer solution. An aluminum rotating disk (220 mm in diameter with a thickness of 10 mm) attached to a laboratory mixer motor (IKA Works Inc.; Wilmington, NC) was used as the ber collector. Rotation speeds of 250 rpm (linear distance 172.7 cm min1), 500 rpm (linear distance 345.4 cm min1) and 1000 rpm (linear distance 690.8 cm min1) were used in this study. The following xed conditions were used in the experiments: (a) applied voltage: 20 kV, (b) distance between the needle tip and the collector: 5.5 cm, (c) collecting time: 1 h (unless stated otherwise), and (d) a constant room temperature of 26 C and a constant relative humidity of 58%. The bers were spun onto 12-by-12 mm glass coverslips (Proscitech, Australia) attached on the edge of the rotating disk using a piece of double-sided tape (3M; St Paul, MN). To determine the optimal electrospinning parameters such as ber alignment and ber density for neuronal outgrowth, different electrospinning conditions were tested (table 1). The diameters of the bers were measured using Scion Image software.
3

SEM was conducted using a Hitachi S-4700 eld emission scanning electron microscope at an accelerating voltage of 1 kV. Each sample was coated with a 10 nm thick layer of platinum/palladium by a Hummer 6.2 sputter coater (Anatech Ltd. Denver, NC). 2.3. Fiber alignment quantication To quantify the alignment of bers, the orientation of 50 bers on each image (from three images captured from independently fabricated samples) was measured using Scion Image. A reference line was drawn along the central orientation and the angle formed between the line and each ber was calculated (gure 2). Each angle was placed into a data bin of 2 , so that all angles between 0 and 2 were placed in one bin, all angles between 2 and 4 were placed in another bin, and so on (Biran et al 2003). Angles ranged from 90 to 90 with 0 being parallel to the reference line. Thus, 150 bers were used to generate the ber alignment graph for each condition. Fast Fourier transform (FFT) was also used to characterize the alignment of the bers. The FFT function converts information present in the original image from the real space into the mathematically dened frequency space (Alexander

J. Neural Eng. 6 (2009) 016001

H B Wang et al

2.6. Cell culture E9 chick DRGs were isolated in accordance with procedures approved by the Institutional Animal Care and Use Committee (IACUC) at Michigan Technological University. The ganglia were divided into halves and placed into a sterile microcentrifuge tube with 200 l of Hanks balanced salt solution (HBSS) (Mediatech, Herndon, VA) and centrifuged at 2000 rpm for 2 min. The supernatant HBSS was removed, and the DRGs were re-suspended into 90 l of neurobasal medium supplemented with L-glutamine, penicillin/streptomycin, and B-27 serum free supplement (Invitrogen, Carlsbad, CA). The suspension was then placed onto crossing bers and highly aligned PLLA bers generated with optimal working conditions. To determine how ber density affected neurite outgrowth, the spinning time was set at 0.5 h and 2 h, which gave signicantly different densities. 0.5 h produced lowdensity bers and 2 h produced high-density bers. DRGs were allowed to attach onto three ber samples (both 0.5 h and 2 h) created from independently fabricated samples that were not coated with any neuronal adherent proteins or serum solutions for approximately 12 h within a tissue culture incubator (37 C, 5% CO2). Then another 2 ml of neurobasal medium was added with a nal concentration of 50 ng ml1 of nerve growth factor (NGF) (Calbiochem, La Jolla, CA). The DRGs were then incubated for 5 days with fresh medium being exchanged every 48 h. Culture experiments were repeated twice to conrm initial results. Schwann cells (SC) were isolated in accordance with procedures approved by the Institutional Animal Care and Use Committee (IACUC) at Johns Hopkins University. Highly puried cultures were obtained from the sciatic nerves of adult female Fischer-344 rats (Charles River Laboratories) as described previously (Hurtado et al 2006) following Morrisseys protocol (Morrissey et al 1991). Dissociated SCs were cultured on poly-L-lysine coated tissue culture dishes in D10 medium (DMEM (Invitrogen, Carlsbad, CA) with 10% fetal bovine serum (Hyclone, Logan, UT) and 0.1% Gentamicin (Invitrogen)) supplemented with the mitogens bovine pituitary extract (2 mg ml1), forskolin (0.8 g ml1), and heregulin (2.5 nM). The solution of heregulin is a previously described (Levi et al 1995) modication of the original protocol (Morrissey et al 1991). To determine the purity of the SCs used, samples of the harvested cells were plated onto culture dishes, cultured for 3 h, stained for S100, and then coverslipped with Citiuor (UKC Chemical Laboratory, Canterbury, UK) with 100 m Hoechst nuclear dye (Sigma, St Louis, MO) to compare numbers of S100positive cells with Hoechst-labeled cells. The purity of the SCs used was 9598%. At early passage, SCs were transduced overnight with a lentiviral vector encoding green uorescent protein (GFP) (Naldini et al 1996) at a multiplicity of infection of 30. The production of the lentiviral vectors has been previously described (Blits et al 2005). GFP expression was controlled by the cytomegalovirus promoter and enhanced with the woodchuck post-transcriptional regulatory element (Loeb et al 1999). The transduction efciency in the SC cultures was >99%. The transduced SCs were further cultured until passage 4 and then collected (Morrissey et al
4

Figure 3. Density measurement. Density (d) is dened as the number of bers along the reference line (Lr2) divided by the length of Lr2.

et al 2006). The resulting FFT output image reects the degree of ber alignment present in the selected area (Ayres et al 2006). A square region of 512 512 pixels on the SEM image was randomly selected and processed for FFT using Scion Image Beta 4.0.3 (Scion Corporation; Frederick, MD). 2.4. Fiber density measurements A new technique for quantifying ber density was developed. Six SEM images (two images from three independently fabricated samples) were selected for each condition. As shown in gure 3, a reference line perpendicular to the majority of the bers was drawn on the SEM image, and the number of bers along the reference line was counted. The density was averaged, and it is presented as the number of bers per mm for each condition. 2.5. Creation of crossed ber specimens and measurement of ber crossing rate To evaluate how crossing bers affect neurite outgrowth, a layer of crossing bers were electrospun onto aligned bers. Highly aligned bers were generated for 10 min using a 22G insulated, sharp-tip needle and using the parameters stated previously (1000 rpm rotation speed for the collection disk and 2 ml h1 syringe pump ow rate). The coverslips were removed from the rotation wheel and reattached after rotating the specimens 45 . Electrospun bers were placed onto the sample for 3 min using the same parameters specied above. Three batches of crossed ber specimens were created and imaged using scanning electron microscopy presented in section 2.2. Within each batch, three images were captured and the number of aligned and crossed bers counted within a selected area (1 mm by 1 mm). The ber crossing rate equation was used to determine the degree of ber crossing within the specimen equation (1): Fiber crossing rate = Number of crossing bers/total number of bers. (1)

J. Neural Eng. 6 (2009) 016001

H B Wang et al

1991) for plating onto three high- and three low-density ber samples with a cell density of 150 000 cells/100 l D-10 medium. SCs were allowed to attach onto three ber samples of independently fabricated samples for approximately 12 h within a tissue culture incubator (37 C, 5% CO2), after which another 2 ml of D-10 medium was added. The SCs were then incubated within the ber scaffolds for 2 days. Experiments were repeated to conrm initial results (N = 2). 2.7. Immunocytochemistry

(A)

(B)

After 5 days in culture, DRGs were xed in a PBS (10 mM phosphate, 150 mM sodium chloride) solution containing 4% (wt/volume) paraformaldehyde (Sigma-Aldrich; St Louis, MO) for 30 min. DRGs were washed three times with PBS, and then blocked with a PBS solution containing 1% normal goat serum (Chemicon, Temecula, CA), 2% non-fat dry milk (TVC Inc.; Brevard, NC) and 0.05% triton X-100 (EMD Chemicals, Gibbstown, NJ) for 15 min. After washing the specimens three times with PBS, DRGs were incubated (37 , 5% CO2) with rabbit anti-neurolament primary antibody (1:200 dilution, Chemicon, Temecula, CA) for 1 h, washed three times with PBS; specimens were incubated with an Alexa Fluor 488 goat anti-rabbit secondary antibody (Invitrogen; Carlsbad, CA) for another hour, and washed three times with PBS. The SCs were washed three times with PBS before uorescence microscopy. Both DRGs and SCs were imaged using a Zeiss Axiovert 200 M microscope equipped with an AxioCam uorescence camera. Zeiss lter set 10 was used where uorescent dyes with excitation wavelengths between 450 and 490 nm and emission wavelengths between 515 and 565 nm are analyzed. Using this lter set, autouorescence from the electrospun bers was not observed. 2.8. Quantication of the density of neurite outgrowth DRGs were placed onto three independently fabricated ber samples for each ber condition (low density and high density) in two separate instances (N = 2). When the neurite outgrowth was longer than 1.5 times the DRG explant diameter, the DRG was considered vital (Kim et al 2005) and was selected for subsequent analysis. Eight DRG explants were selected from the low-density bers (electrospun for 0.5 h), and seven from the high-density bers (electrospun for 2 h). To distinguish neurite extension from the background of the uorescent images, procedures were used similar to those published elsewhere (Bilsland et al 1999, Deister and Schmidt 2006). The micrographs were processed using a processing technique that involved using Adobe R Photoshop CS3. DRGs were removed from the images, as only neurite outgrowth was targeted for measurement. Neurite density was measured by nding the percentage of uorescence in the extension area after neurolament staining. Images were imported into Scion Image software, and thresholding was used to identify the uorescent pixels. Images were then converted to binary images composed solely of white and black pixels, black indicating uorescence and white indicating the background. Using the measure function within Scion Image, the total area (Ao; includes DRG and its neurites), the average gray value of
5

(C)

Figure 4. Diagram of axonal density measurement. (A) Original neurolament-stained image. (B) Photoshop processed image. (C) Measured binary image.

the total area (o) and the area of the removed DRG (Ad) were determined. The gray value of a black pixel is 255 and the gray value of a white pixel is 0. The number of black pixels (x1) in the total area can be determined using equation (2). The neurite outgrowth area (A1) was obtained by subtracting the area of the DRG (Ad) from the total area (A0), as shown in equation (3). The percentage of black pixels (% uorescence), which indicates the density of neurite outgrowth, can be determined by taking the number of black pixels (x1) divided into the area of neurite outgrowth (A1) (equation (4)). The original image, processed image and the measured binary image are shown in gures 4(A), (B) and (C), respectively: 0 A0 255 A1 = A0 Ad x1 100. % uorescence = A1 x1 = 2.9. Quantication of the length of neurite outgrowth The same DRGs mentioned above were selected for neurite outgrowth measurement. Ten of the longest neurites from each side of the DRG explants were measured and averaged to quantify neurite length under different density conditions. The perimeter of each DRG was marked and the length of the neurite was measured from the tip of the neurite to the marked perimeter using Scion Image. (2) (3) (4)

J. Neural Eng. 6 (2009) 016001

H B Wang et al

2.10. Statistical analysis Statistical analyses were performed using JMP IN software (Release 5.1.2; SAS; Cary, NC). A one-way ANOVA was run rst to determine statistical differences between groups in ber density (N = 6), neurite density measurement (N = 8 for low density and N = 7 for high density) and neurite length (N = 160 for low density and N = 140 for high density). For those that showed differences in ANOVA, post-hoc TukeyKramer HSD tests were used to compare all pairs individually. The Brown Forsythe test was run to determine statistical differences in ber alignment (N = 150). A value of P < 0.05 was considered to be statistically signicant.

3.3. The effect of a modied needle on the efciency Because bers are being developed for neural tissue engineering applications, the alignment of bers is critical for directing neurite outgrowth in a particular direction optimally. Our results suggest that if the electrospinning efciency can be improved, a 22G sharp-tip needle would be the best choice to generate the highest degree of alignment. Thus we hypothesized that increasing the intensity of the electric eld on the tip of the needle may increase the efciency of electrospinning. Based on this hypothesis, a piece of tubing (Masterex R , Tygon R , Fisher Scientic, Pittsburgh, PA) was used to insulate the 22G needle body, leaving only the needle tip exposed to the electric eld. A SEM image, a FFT image and angle difference analysis (gures 7(A)(C)) show that the bers were still highly aligned; gure 7(D) reveals that the density of the bers generated by the modied 22G needle was much higher than that of the 22G uninsulated needle (P < 0.05). 3.4. The effect of ow rate A 22G insulated, sharp-tip needle was used at ow rates of 2 ml h1 and 4 ml h1 respectively in order to study the effect of the ow rate on ber alignment. Figures 8(A)(C) contain a SEM image, a FFT image and angle difference analysis. The results show that there is no signicant difference with a ow rate of 2 ml h1 or 4 ml h1 either on alignment (gures 8(A)(C)) or ber density (gure 8(D)). However, with the ow rate of 4 ml h1, more residue was generated because the solvent evaporated before the bers had formed completely. Therefore, a ow rate of 2 ml h1 was used for all other studies. Upon optimization of electrospinning parameters, ber scaffolds were created where 84%, 96% and 99% of the bers deviated no more than 2 , 5 or 10 from a reference line parallel to the aligned bers. Fiber diameters varied between 1.2 and 1.6 m under the optimal conditions. 3.5. The effect of crossing bers on neurite outgrowth Embryonic E9 chick DRGs were cultured on the coverslips containing highly aligned bers described above (gure 8(A)) and crossing bers with a crossing rate of 28%. The average crossed angle was 45 to the previous aligned bers as shown in gure 9(A). Figure 9(B) clearly demonstrates that neurite orientation was diverted away from the axis of alignment and in some cases restricted by the crossing bers. 3.6. The effect of ber density on neurite outgrowth Embryonic E9 chicken DRGs were selected on low-density (0.5 h) and high-density (2 h) bers to investigate the effect of the PLLA ber density on neurite outgrowth. After 5 days of incubation, the DRGs were stained for neurolaments. Figures 10(A) and (B) show SEM images of the bers and gures 10(C) and (D) show uorescent images of neurolament-stained neurites. The density of bers, axon density and outgrowth length are shown in gures 10(E), (F)
6

3. Results
3.1. The effect of rotation speed The effect of rotation speed of the collector was studied at speeds of 250 rpm, 500 rpm and 1000 rpm. A 22 gauge, sharp-tip needle with an inner diameter of 0.4 mm was used with a ow rate of 4 ml h1. SEM images (gures 5(A)(C)), FFT output images (gures 5(D)(F)) and the angle difference measurements (gures 5(G)(I)) demonstrate the alignment of the bers. FFT images were taken from 512 512 pixel selections of the original SEM images and generated an output image containing pixels with a symmetrical shape. The narrower area of the center parts in FFT output images indicates better ber alignment. Figure 5 shows improved alignment with increased rotation speed. Figures 5(F), (I) show that the best alignment was obtained with a rotation speed of 1000 rpm. Fiber diameters varied between 1.2 and 1.6 m at a rotation speed of 1000 rpm. Even though highly aligned bers were successfully fabricated, the efciency of electrospinning/collection, related to the density of the bers, was very low while using a 22G sharp-tip needle, as shown in gure 5(J). Therefore, it was hypothesized that a needle with a bigger diameter or with a at tip may give higher efciency. 3.2. The effect of needle size and tip shape 20G needles (an inner diameter of 0.6 mm) with a sharp-tip and a at-tip were used to fabricate the bers. The corresponding SEM images and FFT output images are shown in gures 6(A), (B) and (C), (D), respectively. Figures 6(E) and (F) show the angle difference measurements for the at and sharp tip needles, respectively. The rotation speed was xed at 1000 rpm and other conditions were unchanged. Figures 6(D) and (F) demonstrate that a 20G sharp-tip needle generated bers that were more aligned than those bers generated by a 20G attip needle, shown in gures 6(C) and (E), but still not as aligned as the bers generated by a 22G sharp needle (P < 0.05). However, a 20G needle (bigger diameter) produced more bers (improved efciency) than using a 22G needle (gure 6(G)). Since the 22G needle produced highly aligned ber species, other needle types were not investigated.

J. Neural Eng. 6 (2009) 016001

H B Wang et al

Figure 5. Alignment and density of PLLA bers using different collection disk rotation speeds. (AC) SEM images of bers collected with a rotation speed of 250 rpm (A), 500 rpm (B), and 1000 rpm (C) (scale bar = 50 m). (D)(F) FFT images from 512 512 pixel selection from images (A)(C), respectively. The center area depicts the alignment of the bers. The narrower areas represent the most highly aligned ber specimens. (G)(I) Histograms of angle difference showing the alignment of bers in (A)(C), respectively. The best ber alignment was obtained when using a rotation speed of 1000 rpm as depicted in panel I (P < 0.001). (J) Plot of the ber density collected under different rotation speeds. 7

J. Neural Eng. 6 (2009) 016001

H B Wang et al

Figure 6. Effects of needle size and tip shape on ber alignment and density. SEM images of bers produced using a 20G, at-tip (A) and sharp-tip (B) needle (scale bar = 50 m). (C) and (D) FFT images from 512 512 pixel selection from images of (A) and (B), respectively. (E) and (F) represent the ber alignment analyses. The best ber alignment was obtained with a sharp 20G needle as compared to the at 20G one (P < 0.001). (G) Density plot showing that the 20G needle is more efcient than the 22G needle (P < 0.05).

and (G), respectively. The results suggest that highly aligned bers guide neurite outgrowth in a directed manner. The density of neurites (gure 9(F)) increased with increasing ber density, 65% 14 of the uorescence for the low density and 83% 6 for the high density (P < 0.05). The length of the neurite outgrowth for 0.5 h was 1210 345 m and for 2 h was 1054 249 m (mean standard deviation). Statistical analysis showed no signicant difference in neurite length. 3.7. Schwann cells on PLLA bers Schwann cells display a typical bipolar, swirling morphology when cultured on a poly-L-lysine coated dish (gure 11(A)).
8

Schwann cell growth on a dish is not directed. Within the ber cultures, SCs grew in a directed manner along the bers in both the low (gure 11(B)) and high (gure 11(C)) density samples. More Schwann cells were seen in the higher density ber sample.

4. Discussion
4.1. Alignment of electrospun ber species Aligned, electrospun bers have been fabricated (Teo et al 2005, Pan et al 2006) for drug delivery (Chew et al 2005) or tissue engineering applications (Baker et al 2008, Chew et al 2008, Choi et al 2008). Recently, electrospun ber

J. Neural Eng. 6 (2009) 016001

H B Wang et al

Figure 7. Effects of modied needle on ber density which is related to electrospinning efciency. (A) SEM image of the bers produced with an insulated needle at the ow rate of 4 ml h1 for 1 h. (B) FFT image from 512 512 pixel selection from image (A). (C) Fiber alignment using insulation. (D) Density plot reveals that the density of bers (A) is much higher than that using an uninsulated needle (compared to gure 5(C)) (P < 0.05). Scale bar = 50 m.

structures have shown great promise in directing neurite outgrowth in a particular direction in both in vitro (Corey et al 2007, 2008, Schnell et al 2007, Kim et al 2008) and within in vivo models (Chew et al 2007, Kim et al 2008). While these studies suggest that ber alignment is important in directing neurite outgrowth, further optimization of ber characteristics may be required. For example, optimally aligned ber scaffolds that minimize ber crossing may better direct neurite outgrowth, potentially improving functional recovery and/or decreasing the amount of time that a functional decit exists. Since previous observations in our laboratory and by others (Corey et al 2007, Kim et al 2008) show that randomly oriented bers have the potential to impede axonal outgrowth, the goal of this study was to produce highly aligned electrospun scaffolds. In the present study, electrospinning conditions were investigated in order to produce highly aligned bers from an 8 wt% PLLA solution in chloroform. While maintaining the same voltage, distance between the needle tip and the collector, and the same room temperature and humidity conditions, four electrospinning parameters were varied: the collection disk rotation speed, the inner diameter size of the needle, the shape of the needle tip and the syringe pump ow rate. After a series of optimization tests, it was determined that a collection disk rotation speed of 1000 rpm, a 22G sharp
9

needle and a syringe pump ow rate of 2 ml h1 produced highly aligned ber scaffolds. Furthermore, when chick DRG explants were placed onto the scaffolds, increased neurite density was seen when explants were cultured on high-density aligned ber substrates. Schwann cells placed onto the aligned ber scaffolds exhibited an extended or stretched morphology as seen in other studies (Chew et al 2008). This work shows that at 1000 rpm, the most aligned PLLA electrospun bers were fabricated. The reason behind this result may be due to the fact that a high collection disk rotation rate stretches the bers into an aligned and uniform orientation without ber breakage. At lower rates (250 and 500 rpm) the alignment was lower, and when the rotation speed was higher than 1000 rpm, the collection efciency decreased (data not shown) because bers were diverted away from the collecting disc by air ow from the rotation of the disk. Therefore, fewer bers were attached to the collector. The process of electrospinning has been described previously (Fong et al 1999, Deizel et al 2001, Huang et al 2003) and has revealed that when an electric eld is applied to a needle, the polymer uid forms a conical shape at the tip of the capillary known as the Taylor cone (Fong and Reneker 1999). The driving force, which is related to the intensity of the electric eld, opposes the surface tension

J. Neural Eng. 6 (2009) 016001

H B Wang et al

(A)

(C)

(B)

(D)

Flow Rate of the Syring Pump (ml/hr)

Figure 8. Alignment and density of the bers generated by a 22G insulated, sharp-tip needle at a ow rate of 2 ml h1. (A) SEM image of the bers produced with a ow rate of 2 ml h1. (B) FFT image from 512 512 pixel selection from image (A). (C) Alignment of sample A. (D) Density plot. Neither the alignment (A)(C) nor density (D) is signicantly different when different ow rates were applied. Scale bar = 50 m.

of the polymer uid. Beyond a critical point, the Taylor cone is unstable and a charged polymer jet ejects out and is deposited onto a grounded collector. As the solvent evaporates, the jet solidies to form nano or microbers (Rutledge and Fridrikh 2007). Since electrospinning highly aligned bers is an inherently difcult process that depends on several factors, manipulation and characterization of factors to create highly aligned ber species provides a useful tool for creating such scaffolds. 4.2. Needle insulation improves electrospinning efciency The 22G needle allowed for the creation of highly aligned ber species in comparison to when the 20G needle was used, but the electrospinning efciency (the number of bers collected over time) was lessened. The electrospinning efciency was also enhanced using a novel needle insulation technique that has not been published before. It is postulated that when the entire needle is charged, without insulation, the electric eld surrounds the needle. Electrical forces around the needle pull bers away from the collection disk. Surrounding the needle with insulation dampens the electrical eld around the needle, and thus more bers are deposited onto the collection disk.
10

4.3. Culturing neurons on highly aligned and crossed ber species and assessing directed neurite outgrowth Culture of primary neuronal cells on aligned bers has improved directed outgrowth of neurites (Corey et al 2007, 2008, Kim et al 2008). From these recent studies, ber alignment was assessed using the fast Fourier transform (FFT) followed by computer analysis or by manually measuring the angle difference of each ber from a reference line parallel to the aligned bers. Although more aligned than random or intermediate aligned bers, bers still crossed as shown in the SEM images of these studies. The inclusion of crossed bers within samples is signicant. To better understand how extending neurites might alter neurite outgrowth, a crossed ber scaffold was created. Culturing DRGs on the crossed ber specimens shows that highly aligned ber species are very signicant because the crossing bers changed the neurites orientation and even stopped directing neurite outgrowth in some instances. In this study, FFT analysis was used in concert with manual determination of ber alignment. Fibers generated using our optimized electrospinning parameters produced ber scaffolds where 84%, 96% and 99% of the bers deviated no more than 2 , 5 or 10 from a reference line parallel to the aligned bers, respectively. Direct comparison of ber

J. Neural Eng. 6 (2009) 016001

H B Wang et al

(A)

(B)

Figure 9. Effects of crossing bers on neurite outgrowth. (A) SEM image of crossing bers with a crossing angle of 45 . 28% of the total bers were crossing bers. Scale bar = 50 m. (B) Neurolament-stained image of DRGs cultured for 5 days on crossing ber scaffolds. Neurites extended along the crossed bers and in some instances were impeded by the crossing bers (circled area) Scale bar = 100 m.

alignment from our study with others (Corey et al 2007, 2008) is difcult since alignment in these studies is computed from FFT images and full widthhalf max analysis (FWHM). While FWHM analysis provided insight into the aligned character of ber scaffolds, additional alignment characterization using other approaches provide clearer characterization of ber alignment. Corresponding neurite outgrowth from DRGs was less aligned than the ber alignment (Corey et al 2007) whereas similar when dissociated primary neurons were used (Corey et al 2008). Our culture data also show directed neurite outgrowth from our DRG explants, but neurite outgrowth was not as aligned as ber alignment. This is likely due to the bers separating from the glass coverslip during cell staining and not because of the lack of ber alignment in our created scaffolds. As the ultimate goal is to apply the aligned ber substrates into a three-dimensional model for in vivo study, our laboratory is developing strategies to secure aligned bers in place for in vitro analysis and for incorporation into a threedimensional nerve bridge. Strategies using polymer lms placed onto coverslips prior to electrospinning (Chew et al 2007) is one method being pursued to stabilize ber alignment (Corey et al 2008). However, the use of lms may shield
11

bers from the grounded collecting disk, requiring additional manipulation of electrospinning parameters to achieve highly aligned ber specimens. 4.4. Increased polymer ber density correlates to higher neurite density but unchanged average neurite length None of these recent studies have examined how ber density affects neurite outgrowth. Our laboratory has developed a rational technique for determining ber density where a line is drawn perpendicular to the axis of ber alignment. The number of bers on this line is counted to yield the number of bers per millimeter (gure 3). Additionally, the neurite density was computed using a technique similar to that presented elsewhere (Bilsland et al 1999, Deister and Schmidt 2006). In our method, black pixels (gure 4(C)), which represented neurite outgrowth, were divided by the total number of pixels in the whole neurite area (A1) to calculate the uorescent percentage as the neurite density. Others have used the area of the actual neurite coverage divided by the entire neurite coverage area to calculate neurite density (Shah et al 2004). Here, instead of using the stained area to indicate the neurite density, the uorescent pixels within

J. Neural Eng. 6 (2009) 016001

H B Wang et al

(A)

(B)

(C)

(D)

(E)

(F)

(G)

Collecting Time (hour)

Neurite Density

Neurite Length

Figure 10. Effects of ber density on neurite outgrowth. (A) and (B) SEM images of bers electrospun at 0.5 h and 2 h respectively. (C) and (D) Neurolament- stained images after DRG were grown on the bers for 5 days. (E) and (F) The density of the bers and the density of axonal outgrowth on each ber condition (P < 0.05), which showed that the density of neurite outgrowth was proportional to the density of the scaffold. (G) The neurite length on each ber condition which was not signicantly different.

images containing neurites were detected automatically using Scion Image, creating a more accurate representation of neurite density than other techniques. Using these novel techniques, it was determined that neurite density increased with ber density. Thus, this
12

suggests that more bers within an area of scaffold lead to more neurites interacting with the scaffold. When thinking of the electrospun bers, the bers can be thought of as being a ridge, while the space between bers can be thought of as being a groove. The dimensions of grooves (space

J. Neural Eng. 6 (2009) 016001

H B Wang et al

(A)

(B)

(C)

between bers) and ridges (diameter of bers) are critical parameters that should be optimized to efciently guide the greatest population of neurites as possible. The dimensions of grooves and ridges have been previously studied in terms of directing cellular outgrowth. Manwaring et al studied the guidance of meningeal cells on imprinted grooved substrates where grooves varied from 10 nm to 940 nm (Manwaring et al 2004). Their results show that the alignment of meningeal cells increased with increasing surface roughness in which the best alignment was found when the depth of grooves was 940 nm. Similarly, other studies have observed axonal outgrowth along larger lament diameters ranging from 5 to 500 m, the most aligned and robust axonal outgrowth was achieved using the 5 m diameter (Wen et al 2006). In another study, when the guidance diameter was less than 1 m, axons grew or crossed over ridges rather than staying within the grooves (Johansson et al 2006). Fiber diameters within this study were larger than those in other studies (Corey et al 2007, 2008, Kim et al 2008). Neurite crossing seen within these studies involving in vitro DRG cultures may be attributed to smaller ber diameters or the presence of Schwann cells (Corey et al 2007, Kim et al 2008). However, when cultured without other cell types, neurites from motor and sensory neurons primarily grew parallel to ber alignment, but crossing of neurites onto other bers appeared to occur (Corey et al 2008). Taken together, these results suggest that the optimal ber diameter to direct embryonic chick DRG neurite outgrowth may be between 1 and 5 m. In addition to ber density affecting the amount of neurite outgrowth, our Schwann cell culturing results revealed a similar trend in that higher ber densities supported more Schwann cell attachment. While neurite density was affected by the number of bers present, the average neurite length was not signicantly different between the two different ber density scaffolds. The average neurite length mean on the low-density bers was slightly higher than the average neurite length mean on the high-density bers. In the 2 h electrospinning situation, ber structures were layered. However, neurites did have a tendency to follow polymer bers below the layer closest to the DRG (data not shown). Thus, neurite outgrowth in thicker ber specimens may facilitate growth through different planes of bers instead of down the axis of the scaffold. This may be one explanation for the neurite outgrowth on the highdensity scaffold being slightly shorter than neurites on the lower density scaffold. However, further morphological study is necessary to ascertain the precise mechanism by which neurites use the polymer bers to direct their outgrowth. 4.5. Important parameters to consider when designing aligned, electrospun bers for nerve regeneration application Therefore, to produce electrospun scaffolds for nerve regeneration applications, the following geometric factors should be considered when developing an implantable scaffold: ber alignment, ber density and ber diameter. While this study shows that very highly aligned bers direct

Figure 11. Schwann cell morphology on different substrates. (A) Fluorescent image of SCs grown on a poly-L-lysine coated dish, which displays the bipolar, swirling, but undirected growth of SCs. (B) and (C) Fluorescent images of Schwann cells cultured on lowand high-density bers respectively for 2 days. On the bers, Schwann cells grew along the bers in a directed manner. Scale bar = 20 m.

13

J. Neural Eng. 6 (2009) 016001

H B Wang et al

neurite outgrowth, the spacing between bers (width of groove) which is related to ber density also inuences the amount of axonal attachment. The diameter of regenerating axons in relation to the ber diameter is another aspect to consider. In vertebrates, there is a large variation in the size of axons, ranging between less than 1 m and up to 50 m (Rydmark 1981, Matthews 1998). In the present study, the ber diameter was on the same scale as the neurites. Based on these results, the diameter of the ber scaffold should be related to the axonal diameter. Fiber diameters one- to three-fold greater than axonal diameters might be of sufcient diameter for nerve scaffold design.

5. Conclusion
According to these results, highly aligned PLLA bers can be fabricated by carefully manipulating electrospinning parameters. These bers directed the outgrowth of neurites and facilitated the attachment of Schwann cells. Further, increasing ber density increased Schwann cell attachment and supported more neurite extension than when cultured on the low-density scaffolds. Crossing bers seemed to divert neurite outgrowth and in some cases stopped neurite outgrowth from occurring. Future work is focusing on maintaining ber alignment for in vitro and in vivo experimentation. Further, studies on how ber density and ber diameter inuence neurite outgrowth in both in vitro and in vivo systems are also currently under way. The goal is to best construct scaffolds containing aligned, electrospun bers for nerve regeneration applications.

Acknowledgments
This work was supported by grants from the Department of Energy, USA, and Seed and Infrastructure Enhancement Grants from Michigan Technological University. The authors would like to thank Dr P Heiden (MTU, Houghton, MI) for her suggestions on electrospinning, Dr D Clupper (MTU, Houghton, MI) and his student D Jaroch for assistance in designing the electrospinning set up, and Eric Minner for manuscript editing and assistance with statistics.

References
Alexander J K, Fuss B and Colello R J 2006 Electric eld-induced astrocyte alignment directs neurite outgrowth Neuron Glia Biol. 2 93103 Ayres C, Bowlin G L, Henderson S C, Taylor L, Shultz J, Alexander J, Telemeco T A and Simpson D G 2006 Modulation of anisotropy in electrospun tissue engineering scaffolds: analysis of ber alignment by the fast Fourier transform Biomaterials 27 552434 Baker B M, Gee A O, Metter R B, Nathan A S, Marklein R A, Burdick J A and Mauck R L 2008 The potential to improve cell inltration in composite ber-aligned electrospun scaffolds by the selective removal of sacricial bers Biomaterials 29 234858 Balgude A P, Yu X, Szymanski A and Bellamkonda R V 2001 Agarose gel stiffness determines rate of DRG neurite extension in 3D cultures Biomaterials 22 107784 14

Bilsland J, Migby M, Young L and Harper S 1999 A rapid method for semi-quantitative analysis of neurite outgrowth from chick DRG explants using image analysis J. Neurosci. Methods 92 7585 Biran R, Noble M D and Tresco P A 2003 Directed nerve outgrowth is enhanced by engineered glial substrates Exp. Neurol. 184 14152 Blits B, Kitay B M, Farahvar A, Caperton C V, Dietrich W D and Bunge M B 2005 Lentiviral vector-mediated transduction of neural progenitor cells before implantation into injured spinal cord and brain to detect their migration, deliver neurotrophic factors and repair tissue Restor. Neurol. Neurosci. 23 31324 Cai J, Peng X, Nelson K D, Eberhard R and Smith G M 2005 Permeable guidance channels containing microlament scaffolds enhance axon growth and maturation J. Biomed. Mater. Res. A 75 37486 Ceballos D, Navarro X, Dubey N, Wendelschafer-Crabb G, Kennedy W R and Tranquillo R T 1999 Magnetically aligned collagen gel lling a collagen nerve guide improves peripheral nerve regeneration Exp. Neurol. 158 290300 Chauhan N B, Figlewicz H M and Khan T 1999 Carbon laments direct the growth of postlesional plastic axons after spinal cord injury Int. J. Dev. Neurosci. 17 25564 Chew S Y, Mi R, Hoke A and Leong K W 2007 Aligned proteinpolymer composite bers enhance nerve regeneration: a potential tissue-engineering platform Adv. Funct. Mater. 17 128896 Chew S Y, Mi R, Hoke A and Leong K W 2008 The effect of the alignment of electrospun brous scaffolds on Schwann cell maturation Biomaterials 29 65361 Chew S Y, Wen J, Yim E K F and Leong K W 2005 Sustained release of proteins from electrospun biodegradable bers Biomacromolecules 6 201724 Choi J S, Lee S J, Christ G J, Atala A and Yoo J J 2008 The inuence of electrospun aligned poly(epsilon-caprolactone)/collagen nanober meshes on the formation of self-aligned skeletal muscle myotubes Biomaterials 29 2899906 Corey J M, Gertz C C, Wang B S, Birrell L K, Johnson S L, Martin D C and Feldman E L 2008 The design of electrospun PLLA nanober scaffolds compatible with serum-free growth of primary motor and sensory neurons Acta Biomaterial. 4 86375 Corey J M, Lin D Y, Mycek K B, Chen Q, Samuel S, Feldman E L and Martin D C 2007 Aligned electrospun nanobers specify the direction of dorsal root ganglia neurite growth J. Biomed. Mater. Res. A 83 63645 Deister C and Schmidt C E 2006 Optimizing neurotrophic factor combinations for neurite outgrowth J. Neural Eng. 3 1729 Deizel J M, Kleinmeyer J, Harris D and Beck Tan N C 2001 The effect of processing variables on the morphology of electrospun nanobers and textiles Polymer 42 26172 Dodla M C and Bellamkonda R V 2006 Anisotropic scaffolds facilitate enhanced neurite extension in vitro J. Biomed. Mater. Res. A 78 21321 Dodla M C and Bellamkonda R V 2008 Differences between the effect of anisotropic and isotropic laminin and nerve growth factor presenting scaffolds on nerve regeneration across long peripheral nerve gaps Biomaterials 29 3346 Fong H, Chun I and Reneker D H 1999 Beaded nanobers formed during electrospinning Polymer 40 458592 Fong H and Reneker D H 1999 Elastomeric nanobers of styrenebutadienestyrene triblock copolymer J. Polym. Sci B 37 348893 Hadlock T, Sundback C, Hunter D, Cheney M and Vacanti J P 2000 A polymer foam conduit seeded with Schwann cells promotes guided peripheral nerve regeneration Tissue Eng. 6 11927 Huang Z-M, Zhang Y Z, Kotaki M and Ramakrishna S 2003 A review on polymer nanobers by electrospinning and their applications in nanocomposites Compos. Sci. Technol. 63 222353

J. Neural Eng. 6 (2009) 016001

H B Wang et al

Hurtado A, Moon L D F, Maquet V, Blits B, J r me R and eo Oudega M 2006 Poly (D, L-lactic acid) macroporous guidance scaffolds seeded with Schwann cells genetically modied to secrete a bi-functional neurotrophin implanted in the completely transected adult rat thoracic spinal cord Biomaterials 27 43042 Johansson F, Carlberg P, Danielsen N, Montelius L and Kanje M 2006 Axonal outgrowth on nano-imprinted patterns Biomaterials 27 12518 Kim I A, Park S A, Kim Y J, Kim S H, Shin H J, Lee Y J, Kang S G and Shin J W 2005 Effects of mechanical stimuli and microber-based substrate on neurite outgrowth and guidance J. Biosci. Bioeng. 101 1206 Kim Y T, Haftel V K, Kumar S and Bellamkonda R V 2008 The role of aligned polymer ber-based constructs in the bridging of long peripheral nerve gaps Biomaterials 29 311727 Levi A D, Bunge R P, Lofgren J A, Meima L, Hefti F, Nikolics K and Sliwkowski M X 1995 The inuence of heregulins on human Schwann cell proliferation J. Neurosci. 15 132940 Loeb J E, Cordier W S, Harris M E, Weitzman M D and Hope T J 1999 Enhanced expression of transgenes from adeno-associated virus vectors with the woodchuck hepatitis virus posttranscriptional regulatory element: implications for gene therapy Hum. Gene Ther. 10 2295305 Manwaring M E, Walsh J F and Tresco P A 2004 Contact guidance induced organization of extracellular matrix Biomaterials 25 36318 Matthews G G 1998 Generation of nerve action potentials Cellular Physiology of Nerve and Muscle (Malden: Blackwell Science) pp 6194 Morrissey T K, Kleitman N and Bunge R P 1991 Isolation and functional characterization of Schwann cells derived from adult peripheral nerve J. Neurosci. 11 243342 Naldini L, Blomer U, Gage F H, Trono D and Verma I M 1996 Efcient transfer, integration, and sustained long-term expression of the transgene in adult rat brains injected with a lentiviral vector Proc. Natl Acad. Sci. USA 93 113828 Ngo T-T B, Waggoner P J, Romero A A, Nelson K D, Eberhart R C and Smith G M 2003 Poly(L-lactide) microlaments enhance peripheral nerve regeneration across extended nerve lesions J. Neurosci. Res. 72 22738

Pan H, Li L, Hu L and Cui X 2006 Continuous aligned polymer bers produced by a modied electrospinning method Polymer 47 49014 Prang P, Miller R, Eljaouhari A, Heckmann K, Kunz W, Weber T, Faber C, Vroemen M, Bogdahn U and Weidner N 2006 The promotion of oriented axonal regrowth in the injured spinal cord by alginate-based anisotropic capillary hydrogels Biomaterials 27 35609 Rangappa N, Romero A, Nelson K D, Eberhard R C and Smith G M 2000 Laminin-coated poly(L-lactide) laments induce robust neurite growth while providing directional orientation J. Biomed. Mater. Res. 51 62534 Rutledge G C and Fridrikh S V 2007 Formation of bers by electrospinning Adv. Drug Deliv. Rev. 59 138491 Rydmark M 1981 Nodal axon diameter correlates linearly with intermodal axon diameter in spinal roots of the cat Neurosci. Lett. 24 24750 Shah A, Fischer C, Knapp C F and Sisken B F 2004 Quantication of neurite growth parameters in explants cultures using a new image processing program J. Neurosci. Methods 136 12331 Stokols S and Tuszynski M H 2006 Freeze-dried agarose scaffolds with unaxial channels stimulate and guide linear axonal growth following spinal cord injury Biomaterials 27 44351 Teo W E, Kotaki M, Mo X M and Ramakrishna S 2005 Porous tubular structures with controlled bre orientation using a modied electrospinning method Nanotechnology 16 91824 Wen X and Tresco P A 2006 Effect of lament diameter and extracellular matrix molecule precoating on neurite outgrowth and Schwann cell behavior on multilament entubulation bridging device in vitro J. Biomed. Mater. Res. A 76 62637 Xu C Y, Inai R, Kotaki M and Ramakrishna S 2004 Aligned biodegradable nanobrous structure: a potential scaffold for blood vessel engineering Biomaterials 25 87786 Yang F, Murugan R, Wang S and Ramakrishna S 2005 Electrospinning of nano/micro scale poly(L-lactic acid) aligned bers and their potential in neural tissue engineering Biomaterials 26 260310 Yang F, Xu C Y, Kotaki M, Wang S and Ramakrishna S 2004 Characterization of neural stem cells on electrospun poly(L-lactic acid) nanobrous scaffold J. Biomater. Sci. Polym. Ed. 15 148397

15